Web Object Encryption and Signing (WOES) BoF
Wednesday, 26 July, 2011, 3:20 pm EDT

Minutes by Karen O'Donoghue

Agenda: www.ietf.org/proceedings/81/agenda/woes.txt
Slides: www.ietf.org/proceedings/81/slides/woes-0.pdf
Jabber: www.ietf.org/jabber/logs/woes/2011-07-26.html

Richard Barnes and Paul Hoffman, as the temporary BoF chairs, called the
meeting to order and presented the traditional IETF Note Well. In agenda
bashing, Pete Resnick asked that the order of the two charter issues be
reversed because he believes the CMS issue is already resolved.

Sean Turner, speaking as AD, reminded the group that security work is hard. He
suggested that the group start from existing stuff, remove stuff, profile,
etc.

Richard Barnes summarized some of the history from the mailing list charter
discussions. Paul asked if there were any other charter comments from the
mailing list that should be highlighted today, and none were identified.

The next agenda item was the CMS charter issue. Joe Hildebrand commented that
CMS has a lot of the security properties that we want. It makes sense to make
use of the security lessons from CMS and carefully edit to remove material
that is obsolete. Pete Resnick felt that this could be stated even stronger.
We want to be basing on CMS semantics but not the data structures as
implemented, focusing on which elements need to appear in a particular profile
and adding things only very judiciously because that would be new work. Eric
Rescorla indicated that he, as the original proposer of the CMS charter text,
meant that we would imitate CMS but not attempt to be cryptographically
compatible. Dave Crocker pointed out that there appear to be at least three
different meaning of "based on" thus far in the discussion. Mike Jones
suggested the language "informed by the lessons learned from CMS". After
further discussion, Peter St. Pierre indicated that we are not tied to a
particular instantiation and will be using well known message security
paradigms. There was consensus that the original text was not intended to
constrain the development. Refinement of the text was moved to the mailing
list.

The discussion then moved to the key format charter item. Mike Jones indicated
that to the extent that it is part of a larger goal of allowing web developers
to have a way of signing and encrypting, you need one. John Bradley indicated
that some have tried to use raw keys. In the interest of interoperability, we
need to do something. Between Eric Resorla and Steve Kent, it was clarified
that we are talking about the algorithms that support PKCS#1. Sean Turner
asked, for the record, why not just use x.509 certificates. Eric Rescorla
indicated that you need to assume you are operating in a non-cert environment.
Bob Morgan said we have been attempting to use certificates in Shibboleth/SAML
for awhile now and it causes endless problems. Further discussion clarified
that how you generate trust in the certificates is outside the scope of the
effort. Public keys are required, but private keys, while convenient, are not
absolutely necessary. Finally, we need to represent asymmetric keys and be
able to convey symmetric ones.

Sean Turner, as AD, indicated that obviously there are folks who are willing
to write drafts for a potential working group. He then asked if there are
folks who are willing to review these drafts, and a fair number of people
raised their hands.

Paul Hoffman asked if there was anything else to discuss today. Eric Rescorla
asked that the working group name be changed from WOES to JOES. Further
charter and working group name discussions were moved to the mailing list, and
the meeting was adjourned.